audiopod-mcp

# AudioPod MCP Server

Hosted [Model Context Protocol](https://modelcontextprotocol.io) server for **AudioPod's audio AI**. Give any MCP-capable agent — Claude Desktop, Claude Code, Cursor, Continue, Cline, Goose, Codex — audio tools over a remote endpoint. No install, no local models, no GPU.

| | |
|---|---|
| **Full endpoint** | `https://mcp.audiopod.ai` — 18 tools: transcription, stems, noise removal, speakers, MIDI, conversion, speech, voices and music |
| **Core endpoint** | `https://mcp.audiopod.ai/core` — 9 tools that work on audio you already have; none creates new audio |
| **Transport** | Streamable-HTTP (MCP `2025-06-18`) |
| **Auth** | OAuth 2.1 — your client opens a sign-in window on first use. For CI / headless use, send `X-API-Key: ap_*` instead ([get a free key](https://audiopod.ai/dashboard/account/api-keys)) |
| **Docs** | https://docs.audiopod.ai/agents · MCP reference https://docs.audiopod.ai/sdks/mcp |
| **Discovery** | [`/.well-known/mcp/server-card.json`](https://audiopod.ai/.well-known/mcp/server-card.json) |

**Registry names** (official [MCP Registry](https://registry.modelcontextprotocol.io)): `ai.audiopod/audiopod` (full, [`server.json`](./server.json)) and `ai.audiopod/audiopod-core` (core, [`server-core.json`](./server-core.json)).

**Installing with an agent?** [`llms-install.md`](./llms-install.md) is the unattended-install guide — config for each client, how to verify, and the failures to expect.

**Using a coding agent?** [`AudiopodAI/audiopod-plugins`](https://github.com/AudiopodAI/audiopod-plugins) wraps this server plus fourteen task skills as a one-command install for the major agent CLIs and editors.

**Prefer HTTP?** The same capabilities are a REST API, and the public **Getting
Started** collection reaches a real `200` in two requests with no account and no
API key. [More about the collections](https://docs.audiopod.ai/sdks/postman).

[![Run in Postman](https://run.pstmn.io/button.svg)](https://god.gw.postman.com/run-collection/58401244-916ddbda-1a70-4c07-8ccd-fe524703a84b?action=collection%2Ffork&source=rip_markdown&collection-url=entityId%3D58401244-916ddbda-1a70-4c07-8ccd-fe524703a84b%26entityType%3Dcollection%26workspaceId%3D8013472d-5897-47d1-bf5c-92683a027b35&utm_source=audiopod-mcp)

## Tools

| Tool | Endpoint | Does |
|---|---|---|
| `transcribe_audio` | full, core | Transcribe with word-level timestamps, speaker labels and SRT/VTT subtitles |
| `separate_stems` | full, core | Split a track into stems (vocals, drums, bass, …); two-stem mode = karaoke |
| `denoise_audio` | full, core | Remove background noise while preserving voice character |
| `separate_speakers` | full, core | Put each speaker on their own track |
| `convert_media` | full, core | Convert audio/video formats (mp3/wav/flac/ogg/m4a, mp4/mov) |
| `audio_to_midi` | full, core | Convert a recording to MIDI |
| `check_job_status` | full, core | Poll the status/result of a long-running job (free) |
| `get_account` | full, core | Check account and credits |
| `get_profile` | full, core | Identify the connected account |
| `text_to_speech` | full | Speech in 200+ languages, 500+ voices and custom clones |
| `clone_voice` | full | Clone a voice (200+ languages) from a 5–30s reference clip, with consent |
| `change_voice` | full | Convert a recording to a different target voice |
| `generate_music` | full | Songs, instrumentals, rap, or vocal stems from a text prompt |
| `compose_music` | full | Compose a song score |
| `revise_music_score` | full | Revise a song score |
| `record_music_score` | full | Record a song score to audio |
| `get_music_score` | full | Read a song score |
| `edit_music_score` | full | Deprecated alias of `revise_music_score` |

Long-running tools return a `job_id`; poll with `check_job_status`.

## Add it

**Claude Code** (OAuth sign-in on first use; swap in `/core` for the core endpoint)
```bash
claude mcp add --transport http audiopod https://mcp.audiopod.ai --scope user
```

**Claude Desktop / Cursor / Continue / Cline** — add to your `mcpServers` config:
```json
{
  "mcpServers": {
    "audiopod": { "url": "https://mcp.audiopod.ai" }
  }
}
```

**Headless / CI** — skip OAuth with an API key header:
```json
{
  "mcpServers": {
    "audiopod": {
      "url": "https://mcp.audiopod.ai",
      "headers": { "X-API-Key": "ap_YOUR_KEY" }
    }
  }
}
```

## Try it (raw JSON-RPC)

```bash
# List tools
curl -s https://mcp.audiopod.ai \
  -H "X-API-Key: ap_YOUR_KEY" -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

# Text to speech
curl -s https://mcp.audiopod.ai \
  -H "X-API-Key: ap_YOUR_KEY" -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/call",
       "params":{"name":"text_to_speech","arguments":{"text":"Hello from my agent."}}}'
```

## Building a startup on this?

Apply to **[AudioPod for Startups](https://audiopod.ai/startups)** — free Pro for 3 months plus developer API credits for eligible early-stage teams.

## Links

- Website — https://audiopod.ai
- API + SDK docs — https://docs.audiopod.ai
- Agent guide — https://docs.audiopod.ai/agents
- MCP setup guide — https://docs.audiopod.ai/sdks/mcp

Three manifests live here and they are not interchangeable: `server.json` and
`server-core.json` are the official registry entries (registry schema) for the
full and core endpoints, `server-card.json` is the MCP server
card — a reference copy, since the live card is served at the Discovery URL above.

## License

MIT — see [LICENSE](./LICENSE). (Covers this listing/README; the AudioPod service itself is governed by the [AudioPod Terms](https://audiopod.ai/terms).)